ChatGPT UX. Your models. Your data.
LibreChat is the open-source ChatGPT clone. Multi-model support (OpenAI, Anthropic, Google, Mistral, local LLMs), modern UI, file uploads, agents. We host it on warp-grade Google Cloud with optional GPU.
See plans from $25/moEverything in the box.
Latest LibreChat + every model provider + GPU options.
LibreChat (latest)
The open-source ChatGPT clone. Modern UI, conversation branching, custom presets, agents, file uploads, code interpreter. Self-hosted.
Multi-provider support
OpenAI, Anthropic, Google (Gemini), Mistral, Groq, Ollama (local), and any OpenAI-compatible API. Switch models per conversation.
Local model support
Connect to your own Ollama, vLLM, or LM Studio instance. Run Mistral, Llama, DeepSeek, Qwen, etc. — all from one interface.
Warp-grade compute
5,000 IOPS Hyperdisk. For local LLM hosting, add a T4 or L4 GPU. Both available on Premium and Enterprise tiers.
Daisy AI included
Daisy is built on LibreChat. You get the same AI concierge experience on every other Leapjuice app — and you can chat with her on this same instance.
Free n8n
Build "when a new LibreChat conversation is created, log it to a Google Sheet" workflows. Free with annual.
Built for the multi-model era.
Teams replacing ChatGPT Team
Same modern UI, support for every model provider, full data ownership. At 25 users, you save $750/year vs ChatGPT Team. At 200 users, you save $30k+/year.
Privacy-conscious professionals
Lawyers, doctors, journalists, consultants who can't send confidential data to OpenAI. LibreChat gives you the ChatGPT UX with full data sovereignty.
Developers and researchers
Compare outputs from GPT-4, Claude, Gemini, Llama, Mistral in the same interface. Use the API key management to swap models per experiment.
ChatGPT is $20-$25/user/mo. LibreChat on Leapjuice is a flat fee with unlimited users. Same OpenAI/Anthropic model access. Your conversations stay on your server.
Open WebUI is a great alternative (we also host it). LibreChat has a more polished UI, better agent support, and a larger plugin ecosystem. Same general capability.
LibreChat requires Node.js 18+, a separate API server, careful configuration of model providers, and ongoing maintenance. We handle it all on warp-grade infrastructure.
Pick a plan.
- 4 vCPU + T4 GPU cpu
- 16 GB ram
- 500 GB ssd
- 10 TB traffic
- 1 instance, 25 users sites
- 8 vCPU + L4 GPU cpu
- 32 GB ram
- 1 TB ssd
- Unlimited traffic
- 1 instance, unlimited users sites
Questions, answered.
Do I need to bring my own API keys?
Yes. You provide the API keys for the models you want to use (OpenAI, Anthropic, Google, Mistral, etc.). You can also run local models (Ollama, vLLM) with no external API needed.
Can I use local models on the same instance?
Yes. Install Ollama or vLLM on the instance, configure LibreChat to use it as a custom endpoint. The same UI then gives you access to both cloud models (OpenAI) and local models (Llama) — switch per conversation.
What GPU do I need for local models?
For 7B-13B parameter models, an NVIDIA T4 (16GB VRAM) is enough. For 30B-70B models, you want an L4 (24GB) or A100. We offer T4 on Premium and L4 on Enterprise.
Is my conversation data private?
Yes. LibreChat stores conversations in MongoDB on your instance. API calls go directly from your server to OpenAI/Anthropic (not through us). We can't see your conversations.
Can I share conversations with my team?
Yes. LibreChat has a shared conversations feature. You can also set up "agents" — custom system prompts that your team can use as starting points.